> mcp.connect()

The MCP Server for AI Database Access

Connect Claude, Cursor, and VS Code to your database. 40+ MCP tools. Ontology-powered. Free forever.

.mcp.json

{
  "mcpServers": {
    "ekaya": {
      "url": "https://your.internal.service/mcp"
    }
  }
}

40+

MCP Tools

query, execute, schema, ontology

DDL

Full Execute

CREATE, INSERT, ALTER, migrations

Ontology Co-pilot

answers questions, enriches metadata

Free Forever

open source, zero config, one URL

Get started in minutes

From zero to MCP-connected database in a few clicks

Connect your database

Add your PostgreSQL connection string. Ekaya supports dozens of datasources via adapters.

→

Select your schema

Choose which tables and columns to expose. Exclude PII or sensitive data from the ontology.

→

Build the ontology

Ekaya analyzes your schema and generates a semantic layer. Your MCP client refines it automatically.

→

Add the MCP URL

One line in your AI tool config. Works with Claude Code, VS Code, Cursor, and any MCP client.

What developers are building

Ekaya turns your AI coding assistant into a database-aware development partner

claude-code

# Analyze CSV headers and sample data

# Design a normalized schema

execute(sql='

CREATE TABLE product_metrics (

id SERIAL PRIMARY KEY,

product_id TEXT NOT NULL,

metric_date DATE NOT NULL,

revenue_cents BIGINT,

units_sold INTEGER

)

✓ Table created

# Now writing ETL pipeline...

# Inserting 12,847 rows from CSV

execute(sql='INSERT INTO ...')

✓ 12,847 rows inserted

Data Engineering

CSVs to production schema in minutes

Hand Claude Code a directory of CSVs and let it design a normalized database schema, create the tables via the execute tool, then write and run the ETL pipeline to ingest the data — all in one session.

▸Analyze source files, infer types and relationships
▸CREATE TABLE, INSERT, ALTER — full DDL/DML access
▸Validate with queries before committing

Product Development

Debug your data layer without leaving code

Investigating a bug in your repository layer? Claude Code can cross-reference your ORM models with the actual database state via Ekaya's MCP tools — checking row counts, verifying constraints, and sampling data without you switching to a SQL client.

▸Query live data while reading application code
▸Validate ORM mappings against actual schema
▸EXPLAIN ANALYZE to diagnose slow queries

claude-code

# User reports duplicate transactions

# Check the actual data...

query(sql='

SELECT transaction_id, COUNT(*)

FROM billing_transactions

GROUP BY transaction_id

HAVING COUNT(*) > 1

→ 3 duplicate transaction_ids found

explain_query(sql='...')

→ Missing UNIQUE constraint on

transaction_id column

✓ Root cause identified

claude-code

# "Add a user notifications feature"

get_schema()

→ 30 tables, reviewing relationships...

execute(sql='

CREATE TABLE user_notifications (

id BIGSERIAL PRIMARY KEY,

user_id TEXT NOT NULL

REFERENCES users(user_id),

notification_type TEXT NOT NULL,

payload JSONB,

read_at TIMESTAMPTZ,

created_at TIMESTAMPTZ DEFAULT NOW()

)

✓ Table + migration created

# Writing Go model + repository...

Feature Development

Schema and migrations from a single prompt

Ask Claude Code to add a feature and it will inspect the existing schema via get_schema, design new tables that follow your conventions, create them via execute, then write the application code and migration files — all in one flow.

▸Understands your existing schema conventions
▸Creates tables, indexes, and constraints
▸Generates migration files alongside application code

Ontology Co-pilot

AI answers its own ontology questions

When Ekaya extracts your schema, it generates questions about business rules, relationships, and terminology. Your MCP client can read these questions, research the codebase, and submit answers directly — no human copy-paste required.

▸Claude Code reads ontology questions via MCP
▸Researches your codebase for the answers
▸Updates column descriptions, enums, and business rules automatically

claude-code

# Claude Code reads a pending ontology question:

list_ontology_questions(status='pending')

→ "What does offer_type integer map to?"

# Researches the Go codebase...

# Found in offers/types.go:12

update_column(

table='offers', column='offer_type',

enum_values=['1 - Video Call',

'2 - Audio Call', '3 - Chat']

)

resolve_ontology_question(

question_id='...',

notes='Found in offers/types.go:12'

)

✓ Ontology updated

Open Source & Zero Lock-in

True open source with Apache 2.0 license — your insurance policy against vendor risk

Complete Source Access

▸Apache 2.0 License: Full commercial use rights with no restrictions
▸Public Repository: All code on GitHub, fully auditable
▸No Hidden Components: Everything needed to run in production included
▸Fork-Friendly: Take the project in your own direction if needed

Self-Hosting Freedom

▸Behind Your Firewall: Deploy in your VPC with complete control
▸Air-Gapped Support: Run in isolated environments without internet
▸Custom Applications: Deploy your own applications at no cost
▸Your Infrastructure: Cloud Run, Kubernetes, bare metal — your choice

The MCP Server is the foundation of Ekaya's AI Data Readiness Framework — the starting point at Stage 1 and the infrastructure that powers every stage through Stage 5.

Explore the AI Data Readiness Framework

Start Building Today

Get Started for Free Take the Assessment